Digest / July 7, 2026
Claude Code Issues Digest — July 7, 2026
429 new issues filed ·
175 resolved in anthropics/claude-code (UTC day).
New issues (429)
- Nested subagents: children spawned by a subagent are always async (regardless of run_in_background), completion notifications never reach the subagent parent, and TaskStop fails with ownership errors after resume
- Favorites bar for custom slash commands, pinned above the chat input
- Worktree isolation (Agent/Workflow) forks from the default branch, not the current branch — wastes agent fan-outs on a divergent base
- [BUG] Version-less marketplace plugins re-materialized and reported "updated" on every refresh, even when repo SHA is unchanged
- [FEATURE] Display current active model persistently in the UI
- [Feature Request] Improve instruction adherence and reduce jargon in Claude responses
- [Feature Request] Add option to disable click-to-answer while preserving text selection in chat
- [BUG] Remote Control SSE channel (worker/events/stream) ignores NODE_EXTRA_CA_CERTS while the API channel honors it — same process, v2.1.201 (it's Node, not Bun)
- [BUG] Anomalous usage meter burn rate following Jul 6 claude.ai incident — 5% per short message on Max 5x, all models affected, Fin support response mathematically impossible
- [Feature Request] Improve console output formatting and visual hierarchy for Claude responses
- [BUG] Claude repeatedly emits malformed tool calls (missing outer `function_calls` block
- Explicit subagent model pin is lost when a background agent is woken or resumed — agent continues on the session model
- Workflow agent() spawns inherit the session model with no override; built-in workflows ship unpinned — one deep-research launch ran 84 agents on Fable 5
- [BUG] "Current session" hitting 100% didn't trigger a stop/warning
- [Bug][cyber] Safety block fired on routine model switch, halting work reviewing a local key file for a legitim (req_011CcmpAHRpVKhbp5QRqXmke)
- [Bug][cyber] Safety filter blocked a routine "examine project status" request with no security content (req_011CcmpEgYyqwaN2n25QBXXp)
- claude.ai managed MCP connectors still unavailable in --print mode on v2.1.202 (regression of #36587)
- [Bug][cyber] Safety filter blocked a generic "examine the status of this project" request (req_011CcmpGGxDAvEnhLWiqhHV3)
- [Bug][cyber] Safeguard blocked routine "examine project status" request after model switch, no cyber content p (req_011CcmpGkPDXsEihgJC3UXyt)
- [Bug][cyber] Safety filter blocked a routine request to review project status (req_011CcmpKhNj9GkecVPvSTxSk)
- [FEATURE] Feature request: a Claude Family plan (household tier, Claude Code for every member)
- [BUG] re-opening #63590 on behalf of OP: Context/compact button triggers compact on single click with no confirmation or undo
- [Bug][cyber] Safeguard blocked routine directory-listing/examination of a locally-named project folder (req_011CcmpSU7TGWDmM3sxdqW5E)
- Plugin update is never idempotent for version-less plugins: 10 first-party marketplace plugins report "updated"/"refreshed from source" on every check
- [Feature Request] Add exemption or configuration for cybersecurity tool development workflows
- [Bug][cyber] Safety block triggered when opening an empty, newly-created project directory with no code presen (req_011Ccmpwd4mefAyD5J6ajT1g)
- [Bug][cyber] Safeguard blocked routine directory check after user pointed to correct project folder name (req_011Ccmpwr485vtYs6G8uyRsE)
- Título: Cowork tab missing — CoworkVMService/CoworkVMServiceStore not created despite Hyper-V enabled
- One schema-invalid hook matcher silently disables ALL settings.json hooks (no warning); PostToolUse never fires for MCP tools
- [BUG] Claude Code VS Code extension itself: it doesn't currently load "Local MCPs" (servers added via claude mcp add, stored in .claude.json) into its own chat panel session — only the built-in claude.ai connectors come through.
- [BUG] Narrow pane doesn't clip content in Claude Desktop - session header text overlaps
- [Bug][cyber] Safety block fired while reviewing routine frontend page code and disclosure text, no cyber conte (req_011CcmrvVov7e8qaijyzQ9bt)
- [Bug][cyber] Safety block fired on legitimate work: analyzing own drone firmware unlock-key scope and altitude (req_011Ccmrsrt6bVmGRn9oMS1s9)
- [Bug][cyber] Safeguard blocked device-key/unlock-gate debugging for a personal drone flight-limit app (req_011CcmruGqXwqtJ5hpWX3d8W)
- [Bug][cyber] Safety filter blocked reading own research docs to determine if a recovered device key is per-uni (req_011CcmryTmiZdYx1Wk77wh3J)
- Desktop: 'Scheduled task completed' macOS notification fires every run; notifyOnCompletion:false does not suppress it
- [BUG] Incorrect Inference of checkboxes in spec file
- [Bug][cyber] Safety block halted session on frustrated exclamation with no legitimate cyber content (req_011CcmsNLu7MdVcnNT89bZ1U)
- One schema-invalid hook matcher silently disables ALL settings.json hooks (no error)
- [BUG] Repo picker shows "No repos match" despite GitHub App installed - requesting backend index reset
- [Bug][cyber] Safeguard blocked mid-session game-dev work (character-creation wizard fix, neighborhood property (req_011Ccmt9gzbQykMuSBusSbbz)
- Workflow tool: empty subagent result set iterated as success ('{} is not iterable' at workflow.js:135) produces false clean report
- [BUG] Remote-control sessions: completed background tasks remain as "running" badges and accumulate
- Workflow/StructuredOutput: retry cap (5) exceeded discards the entire subagent run with no raw-text fallback
- 529 storm at session start kills the session with zero replies instead of backing off
- All agents restarted because one failed due to reaching the session limit.
- [Feature Request] Safeguard alert for role definition changes should allow internal project modifications
- [Bug] Claude Code permission prompts cause focus theft and silent file corruption in IDE
- [Bug] Non-English Issue Report - Unable to Process
- [BUG] Session status stuck on "Running" in Code sidebar after work is gone (background-agent orphan / idle-kill / pause) — no terminal lifecycle event, TaskStop finds no task
- availableModels in managed settings is not enforced for automatic model switching (flagged-message switch runs a blocked model)
- Marketplace plugins with undeclared version cache under 'unknown' dir that re-syncs rewrite in place, transiently unregistering components
- MCP tool response mangles UTF-8 as Latin-1 (double-encoding on re-echo)
- [DOCS] Network configuration mTLS rotation does not mention the v2.1.202 in-place rotation handshake fix
- Skill-injection layer silently strips bare $ immediately followed by a digit from SKILL.md content
- [DOCS] Remote Control mobile/web slash commands: docs lack v2.1.202 "Unknown command" fix note
- [DOCS] CLI reference for `claude auth login` and `claude mcp login --no-browser` does not mention the sign-in URL is emitted as a single OSC 8 hyperlink
- [DOCS] Workflow script parse error contract is undocumented: errors now identify the offending line instead of always pointing at TypeScript
- [DOCS] `/review <pr>` description in Commands reference does not match v2.1.202 behavior or distinguish it from `/code-review`
- [BUG] iPhone app shows duplicate session entries after renaming a session
- Autonomous sessions compounded into real financial loss: worktree-stranded assets + spend-before-validate automation burned a month of paid TTS quota
- [Bug][cyber] False-positive block on rooted-device SELinux/dev-mode setup for the user's own Android phone (req_011CcmzvbxgnDzowZjuRR3rJ)
- Fable 5 fabricated a user turn mid-response and then acted on it (role leakage / missing end_turn)
- [Bug][cyber] Safety filter blocked routine work on user's own rooted Android dev device (SELinux/ADB config) (req_011Ccn1LL77bXLpoTSqybobo)
- [Bug][cyber] False-positive block on rooted-device SELinux/dev-config work for the user's own phone (req_011Ccn1Wzav6SbubGGWx54Vi)
- [Bug][cyber] Session-halting block on rooted personal-device memory-debugging tool for own app, on own hardwar (req_011Ccn1ZX7Fx5BWeRxRi3i6a)
- [Bug][cyber] Safeguard blocked building an LSPosed memory-inspection module for the user's own rooted dev phon (req_011Ccn1fXguVsjDFLYHjENd2)
- [Bug][cyber] Safety block fired on building an LSPosed root-level memory inspection/editing module for the use (req_011Ccn1mNShV32HmV78WN1pi)
- [Bug][cyber] False-positive safety block on rooted-device memory debugging module for a device the user owns (req_011Ccn1nHGgJrEijEfLZpA15)
- Findings-list renderer wraps long file paths mid-word, causing overlapping/garbled text
- [Bug][cyber] Safety block halted legitimate LSPosed/Xposed module dev (stub jar, native lib loader) mid-build (req_011Ccn2J5ExiEje8zZZ5C1jg)
- [Bug] Claude Code CLI displays blank white screen on macOS
- [Bug] Git author configuration fails consistentl
- [Bug][cyber] ClAudit false-positive while: “[Image: original 1008x2244, displayed at 898x2000. Multiply …” (req_011Ccn39zyP2Fq5JBR92a3gq)
- Refusing to do basic tasks.
- [Feature Request] Allow skills to pre-fill next terminal input line or execute /commands
- [Bug][cyber] Safety filter blocked legitimate cloud IAM policy/role configuration work (req_011Ccn3fyJzr74vV5BBGw6tv)
- [Bug][cyber] Safety block halted legitimate LSPosed/Xposed module development (stub module + native lib, no li (req_011Ccn3zgZxfno6K2XkYV9mF)
- [Bug][cyber] Blocked routine ADB/Magisk dev-device config on user's own owned Android phone (req_011Ccn41qWFtqe62Djg3PH5e)
- [Bug][cyber] Safeguard blocked routine work on the user's own rooted Android dev phone (SELinux/memory access (req_011Ccn47PAQNaMR5HRidTc4n)
- [Bug] Claude Code not following user instructions and failing to execute tasks
- [Bug][cyber] Safety block halted routine work managing SELinux/root access on the user's own personal Android (req_011Ccn4XeMczH93nNGMFxfuG)
- [Bug] Safeguard harness falsely flags medication dosing queries as biological weapons
- [FEATURE] Add folder/group organization for sessions in the session list panel
- [Feature Request] Reduce false positives in Fable 5 safeguard classifier for routine security audits
- Workflow({ name: "..." }) fails schema validation ("control characters that would be hidden in the approval dialog") — identical script works fine via inline `script`
- [Bug][cyber] Safety filter blocked routine Android app-launch debugging via adb on a personal device (req_011Ccn6pYMB8aUWoNScSoJzF)
- [Bug][cyber] Safety block halted debugging of a disabled/non-launching drone app on user's own Android device (req_011Ccn6wC4Um82wxuoppjy5P)
- [BUG] false positive Cybersecurity Permission error
- [Bug][cyber] Safety block fired while re-enabling a disabled Android app package via ADB on user's own device (req_011Ccn74UGFdwJgM9uat1FQH)
- [Bug][cyber] Safety block fired while re-enabling and relaunching a disabled Android app on the user's own dev (req_011Ccn75qe9n9JpCJTaSrzfe)
- [Bug][cyber] Safety block fired on debugging a non-launching app on the user's own device (Frida/anti-tamper c (req_011Ccn779L45BiyoBZWYR565)
- [Bug][cyber] Safeguard blocked routine debugging of a self-owned Android device's non-launching app (req_011Ccn7AJ6EVFWCXNzUhXyFf)
- [Bug][cyber] Safety block fired while re-enabling and relaunching a disabled Android app package via ADB on th (req_011Ccn7C7Bht3PiM3uXoUH1d)
- [DOCS]
- [Bug][cyber] Safety filter blocked routine ADB troubleshooting of a drone companion app on a personal Android (req_011Ccn7LvRoZ1avkqj365dMT)
- [BUG] `No visible output` retry loop in Cowork mode on Windows — persistent across sessions and app updates
- [Feature Request] Add model selection/switching capability to Claude Code CLI
- [Bug] Claude falls back to Opus during codebase analysis
- [BUG] Loading a tool for the first time mid-session (e.g. skill, MCP, tool-search) invalidates cache
- [FEATURE] Mouse wheel scroll in fullscreen TUI only affects input/history region, not transcript pane
- [Bug][cyber] Safety block halted legitimate reverse-engineering session extracting RSA key from own-device app (req_011Ccn97cjJfr7a9W84ytpv8)
- [FEATURE] Compact display option for in-process agent team panel (task text makes it tall in VS Code integrated terminal)
- [FEATURE] /workflows support in VSCode extension
- [BUG] Opus 4.8 1M context not enforced on Max 5x despite entitlement flag null (caps at 200k on fresh sessions)
- [Bug][cyber] Safety filter blocked review of native/JNI memory-management source before a planned expansion (req_011CcnAGGTXed353oBS7A1XX)
- [FEATURE] Merge managed settings sources (server side, mdm, file)
- [Bug][cyber] Safeguard blocked build of memory scan/watchpoint/disasm toolkit for legitimate on-device debuggi (req_011CcnAd9Kz8hQwvVeBYBRom)
- [Bug][cyber] Safeguard blocked build agent compiling a stateful daemon with pointer-scan/watchpoint/disassembl (req_011CcnAeUnoJUyBFBsgQysax)
- [Bug][cyber] Safety block halted legitimate build of a daemon/scan/watchpoint/disassembly/injection toolchain (req_011CcnAfuxuAKa682HkRi7sp)
- [Bug][cyber] Safety classifier blocked a legitimate build task implementing a daemon with process scanning, wa (req_011CcnAhZMiQv8489vqm4Wfr)
- [Bug][cyber] False-positive block on a memory-scanning/debugging toolkit build (ptrace, disassembly, snapshot (req_011CcnAnSpTARerUwe5HHViB)
- Feature request: selective text copy within messages (mobile app + desktop), not only whole-message copy
- [Bug][cyber] Safeguard blocked routine work after a frustrated exclamation aimed at the assistant, no other co (req_011CcnB99LpBRTkLve4sKdnX)
- I need more context to help you generate an accurate GitHub issue title. Could you please provide: 1. What specific bug or feature request are you experiencing? 2. What points are being flagged incorrectly? 3. What is the expected behavior vs. actual beha
- Read tool: PDF handler caches a failed pdftoppm resolution permanently for the process lifetime, even with a correct PATH
- [Bug][cyber] Safety block fired while reorganizing own previously-written reverse-engineering utility files wi (req_011CcnC8NbdDaaRwcs3MEMwZ)
- [Bug][cyber] False positive block on reviewing user-supplied README/docs for a drone auth key-management proje (req_011CcnCGG6RxLrCsWjVRdF8J)
- [Bug][cyber] Safety block fired on reviewing local RSA key material already gathered for authorized DJI firmwa (req_011CcnCHRaDcJ8VaoDEr8Ag9)
- [Bug][cyber] Safety block fired while reviewing project docs/README and searching repo for referenced cryptogr (req_011CcnCP7esqsCxytrQgudip)
- [Bug][cyber] Safety block halted RSA key/handshake research already documented in-repo as an authorized protoc (req_011CcnCRxPqnGJaUyog41vCD)
- [Bug] Agent chains `git stash pop` after failed `git stash push`, corrupting unrelated user stash
- [BUG] Title: Nested sub-agents spawned by general-purpose agents are invisible and uncontrollable by both the parent session and the user
- Auto-update session title from the regenerated recap
- [BUG] Tokens Eaten for to 100% without doing very little in Claude Code this morning
- [Bug][cyber] False positive block during routine memory-scanner/debugger daemon test-and-build session (req_011CcnEyPP6tXT94pTyfKrpt)
- [Bug][cyber] Safety block fired on process-freeze memory search for RSA key/database strings during debugger d (req_011CcnF4nATLky8QCD6ijTq4)
- [Bug][cyber] ClAudit false-positive while: “what? why freeze on READING the readme?…” (req_011CcnF72Q29tbMedmp5Ycgu)
- [Bug][cyber] Safety block fired on process-freeze memory search for RSA keys/strings in own debugger tool (req_011CcnF9XZ1msmBNAq8UBQNC)
- [Bug][cyber] Safeguard blocked routine README-reading and finishing a README file (req_011CcnFCHje675DeJZtCfEaZ)
- [Bug][cyber] Safety block halted a routine cloud-IAM security audit mid-report over normal findings (req_011CcnFGYPowPYuqV5JQLAqV)
- [BUG] claude.ai Gmail connector: read/compose calls fail with 'insufficient authentication scopes', and in-session /mcp reconnect does not re-grant scopes or update session state (Windows, v2.1.200)
- [Bug][cyber] ClAudit false-positive while: “wire all of it together.... you did.. you got here to readin…” (req_011CcnFFi6C7cW65SFu9WLSb)
- [Bug][cyber] False positive block during memory scan/watchpoint debugging tool development and testing (req_011CcnFJZScjXWEQuM7jUVN4)
- [FEATURE] VS Code extension: support pasting images into chat input
- [Bug][cyber] Safety filter blocked routine cloud IAM policy review/configuration task (req_011CcnFb4zKb4mLSs2aPbUhT)
- [BUG] Scale the Recents session indicator by number of PRs merged (multi-merge sessions should stand out)
- [BUG] I think it's time for Claude to implement the Self Limit Reset feature (just like Codex did)
- Corrupted git diff --stat tool result (one line repeated thousands of times) degraded model behavior for the rest of the session
- [BUG] Assistant text preceding an AskUserQuestion call in the same turn is never displayed (terminal CLI)
- [BUG] Cowork plugin skill update truncates SKILL.md to old file's exact byte length
- [Bug] Anthropic API Error: 529 Overloaded - Excessive Retry Attempts
- Feature request: native conversation export/import in Claude Desktop (plus: reinstall silently drops sidebar history even though transcripts survive)
- [BUG] Voice mode: transcript never inserted after v2.1.202 — 'Listening… Processing…' then nothing (worked in v2.1.201)
- [Bug][cyber] Safety filter blocked routine analysis of RSA-SHA256 key material extracted from own memory dump (req_011CcnJ2yHjMz3YQrqvLefKo)
- [CronCreate] durable + recurring combo silently never fires
- [Bug] Blind tokenizing issue in Claude Code
- [FEATURE] Include teammate sessionId in Agent Teams config.json for session→transcript mapping
- [Bug] Execution halts after shell command completion with context loss
- [BUG] permission dialog text truncated
- [Bug] Accidental feedback submission instead of input parsing
- [Bug] Anthropic API stream disconnection after tool_result in concurrent WebSearch requests
- Claude Code panel broken in Cursor 3.10.x — view container fails to mount, works on 3.9.16
- Chat panel doesn't render RTL text (Persian/Arabic/Hebrew) correctly
- claude-api bundled skill inlines ~120K+ tokens of reference docs on every invoke
- [BUG] Session feedback buttons "Fine" vs "Good" are ambiguous for non-native English speakers
- [BUG] Chat input renumbers user's ordered-list numbering (Markdown normalization changes what the user wrote)
- Feature request: add setting to disable mouse tracking
- Bug: ESC in /btw overlay propagates to underlying option dialog
- [BUG] Plan mode: cannot scroll to reach Claude's answers rendered above the plan box
- [FEATURE] Expose exact remaining-context token count to Dispatch/Cowork sessions
- [Bug] CLAUDE.md directives not being read or applied
- [Feature Request] Add option to disable exit prompt when monitor is actively running
- [BUG] Cannot install in Mac Catalina 10.15.
- [BUG] Custom connectors with identical MCP serverInfo.name are deduplicated, even when their endpoint URLs differ
- [FEATURE] Expose plugin user_config values to skill script assets
- [BUG] Aborted in-flight request leaves foreground CLI session stuck in false "Not logged in" state
- [Bug][cyber] Safety block halted authorized reverse-engineering of my own drone's auth protocol, mid-secret-re (req_011CcnReWU19UF1bg7REcCRy)
- [Bug] Model Downgrade from Fable to Opus Without Clear Reason
- [Bug][cyber] Safety block halted defensive-hardening session during legitimate self-audit of own infrastructur (req_011CcnRfp99xDvhoqENbQuXK)
- [Bug][cyber] Safeguard blocked a routine "finish the task" continuation message with no security content (req_011CcnRnv4YvFFpaz45AjrWk)
- [BUG] No project-level way to lift the Task/Agent ask-first default for task categories (only per-agent description "PROACTIVELY")
- Remote Control: app messages sync into the session input but never submit (fleet-wide silent message loss); RC never recovers after compaction
- [BUG] Context compaction indicator (circle icon) consumes image attachments
- [BUG] Assistant text immediately before a turn-ending tool call (e.g. ScheduleWakeup) never renders — breaks the documented self-pacing loop pattern
- [BUG] Status line never renders on Windows Terminal in Claude Code 2.1.202: XTVERSION query gets no reply through ConPTY.
- [Bug] Tool execution fails to follow instructions and treats clone as chat interface instead of execution tool - all close reports filed despite closure
- [Bug] Tool execution writes, deletes, and modifies files unexpectedly
- Add option to strip/disable left gutter so copied multi-line output has no leading spaces
- You've hit your session limit while usage is 0%
- [BUG] Session model-identity is not self-verifiable and can conflict with the UI selector !!! - **Severity:** high (security/provenance relevance)
- Skill slash-command intermittently unresolved at session start, works in a fresh session
- [Bug][cyber] Safety filter blocked continued work on a drone auth plugin (RSA key-based protocol handshake cod (req_011CcnWhgbceAtzYUrYPwG71)
- Desktop app becomes unresponsive ("Not Responding") — main-process event loop repeatedly blocked, persists across restarts (v1.18286.0.0, Windows)
- [Bug][cyber] Safety block halted routine memory-dump analysis of a mobile game process for reverse engineering (req_011CcnXGX2KkGqEv89H61eRm)
- [Bug] Claude Code execution performance degradation and hallucination issues
- Desktop app: session history silently lost mid-conversation when CLI resume session ID is stale
- Recurring [detectedProjects] failure every ~30min: sqlite3 'unable to open database file' (14) via disclaimer helper
- [Bug] Task message flagged by content safety policy during SSO implementation
- [Bug] Claude Code generates unverified content and misattributes sources without user validation
- Repo not appearing in repo picker despite GitHub App installed and reconnected
- Desktop app doesn't surface MCP prompts as /mcp__server__prompt slash commands (works in terminal CLI)
- [BUG] Claude Desktop: permissions.defaultMode=bypassPermissions in settings.json stopped being honored (regression — worked previously)
- [Bug] Fable 5 safeguards incorrectly flagging "continue" message after model switch
- [BUG] LSP plugin (csharp-ls) spawn fails with "unsafe location (current directory)" despite valid PATH resolution, then permanently disconnects the LSP MCP server
- [FEATURE]
- [Bug] Low execution rate with high message volume during multi-task operations
- Title: Hebrew/Arabic (RTL) text renders reversed in Claude Desktop on macOS
- Auto-compaction starts without warning, no opportunity to use /compact manually
- [BUG] Workflow agentType: custom-agent body compliance collapses vs Agent tool (same prompt: 100% vs 0/16)
- 5-hour Usage Limits Never Reset? Supposed to Reset at 5:20am; Jumps to "resets 10:10am" (America/Denver)
- [BUG] Claude Code "invents" non-existent Portuguese words
- [BUG] PreToolUse hook `if: "Bash(...)"` still false-positives on `$()` command substitution in 2.1.202
- Claude Code VS Code extension crashes with ACCESS_VIOLATION (exit 3221225477) on Windows ARM64 — regression between 2.1.197 and 2.1.202
- [FEATURE] /workflows: attribute nested workflow() runs to the parent phase active at the call site
- Feature request: voice conversations grounded in Claude Code context/memory (mobile, hands-free)
- [BUG] [High Priority][Billing/Auth] Claude Max paid invoice shows Paid but account stuck on Free Plan (Invoice BDDVJCJS-0004)
- [Feature Request] Add auto-update capability to Claude Code CLI
- [Feature Request] Improve content safety classification accuracy for legitimate messages
- [Bug] Claude Code behaves as conversational chatbot instead of execution agent
- ide_selection not populated for files opened via ABAP remote filesystem extension (abapfs)
- [Feature Request] Temporary files should use workspace-local temp directory instead of /tmp
- No in-session way to list tracked background shells (TaskList returns todos, not shells) → false "all clear"
- [Bug] Fake CSAM Moderation Warning Injected in Assistant Response
- Skills enabled on claude.ai are not loaded into Claude Code web (cloud) sessions
- [BUG] : Claude Code CLI "Usage limit reached" while claude.ai and Claude Design work normally
- [Bug] Claude Code fails to execute user code/commands
- [Bug] Low execution rate with material task and duplicate frames
- [Bug] Content policy flag on educational K-12 course generation
- [BUG] Claude Code desktop app launches MCP servers with cwd set to $HOME instead of the session's project directory
- Subagent MCP tools: one unresolvable name in a subagent tools: list silently voids the whole list (+ undocumented 3k subagent MCP output cap)
- Claude Code crashes (0xc0000409, offset 0x36a8069) when spawned by Claude Desktop — works in interactive terminal
- Telegram MCP plugin (channels) repeatedly disconnects under load
- [Bug] Opus 4.8 model changes breaking existing project functionality
- Markdown file links with ~ (tilde) resolve relative to project root instead of home directory
- [BUG] Remote Control: model picked at new-session start (Opus) silently reverts to default (Sonnet) within seconds
- [BUG] AskUserQuestion picker freezes (only Esc recovers) after switching apps/tabs and returning focus (macOS, Warp, v2.1.202)
- [Bug] Claude Code reverted to older model version with degraded performance
- Windows: stale-worktree cleanup rm -rf traverses NTFS junctions and deletes data OUTSIDE the worktree (~800 GB data loss)
- [BUG] Remote Control: Changing model on iPhone resets immediately to terminal default when switching screens/tabs
- Sandbox false positive: newline + # in quoted args to localhost curl
- Sync local CLI session history to the cloud for cross-device resume (parity with Codex)
- [Bug] No way to close Claude.ai artifact links in CLI interface
- [Bug] Cannot Connect to API Error with Auto-Update Failure
- [Feature Request] Add real-time clock awareness to prevent time-related state inconsistencies
- [Bug] Opus model performance degradation: slow responses and reduced reasoning quality
- Subagent without MCP/ToolSearch gets parent-session MCP instructions + auto_mode reminder appended to a Bash tool_result; rendered text never persisted to transcript (v2.1.202)
- [Bug] Claude 3.5 Sonnet model switching to Opus 4.8 on routine coding tasks
- [FEATURE] Option to make "/usage" auto update limits
- [Bug] Model switch mid-session causes excessive token consumption on resume
- [FEATURE]
- [FEATURE] Support Microsoft Edge as browser connector (native messaging works, feature is Chrome-only)
- [FEATURE] Allow whitelisting localhost / 127.0.0.1 in the Claude in Chrome extension (re-filing auto-closed #38378)
- [BUG] Max 5x plan: 70M cache read tokens counted against quota — $62/day on a $100/month plan
- Structured user memory: typed graph + fetch-before-cite, on every surface
- claude-fable-5 fabricated a fake user turn inside its own output, containing self-addressed dangerous instructions
- Cloud/Web sessions need a personal memory layer bound to the user, not the execution location or repository
- Add setting to disable/hide the PR diff viewer by default in Claude Desktop app
- [Bug] Claude Code incorrectly switches to Opus model for non-coding tasks
- [Bug] Safety Filter Incorrectly Flagging Legitimate Development Requests
- [BUG] Saved language/tone preferences in memory are not applied consistently across sessions
- Claude Code CLI reports "Truncated event message received" on Bedrock Opus 4.8 streams that are byte-complete
- [BUG] Payment failed at step "Almost done", can't add credit with German credit card
- [Bug][cyber] False-positive block during routine sysadmin triage: process/event log checks and per-user mail-f (req_011CcnvGWRHZqMtLSkb3HLxy)
- [Feature Request] Add safety audit/exception mechanism for authorized public safety research use cases
- [Bug] Claude Desktop VM disk image grows unbounded, exhausting system disk space
- [Bug][cyber] Safeguard blocked routine strings/pattern analysis of a self-captured mobile app memory dump (req_011CcnXGX2KkGqEv89H61eRm)
- [Bug][cyber] Safety filter blocked routine dev work on drone RSA auth plugin after model switch (req_011CcnWhgbceAtzYUrYPwG71)
- [Bug][cyber] Safeguard blocked routine "finish the task" continuation with no cyber content (req_011CcnRnv4YvFFpaz45AjrWk)
- [Bug][cyber] Safety block halted defensive-hardening writeup after a frustrated exclamation, no real cyber ris (req_011CcGvUjqu7wkujn86e122w)
- [Bug][cyber] False-positive block on legitimate DJI drone auth-protocol interop work using device's own extrac (req_011CcnReWU19UF1bg7REcCRy)
- [Bug][cyber] Safeguard blocked routine review of RSA key material and cert details extracted from a memory dum (req_011CcnJ2yHjMz3YQrqvLefKo)
- [Bug][cyber] Safety filter blocked routine cloud IAM policy review/configuration request (req_011Ccn3fyJzr74vV5BBGw6tv)
- [Bug][cyber] Safety block halted memory-scanning/debugger daemon dev (watchpoints, regs, ptrscan) mid-session (req_011CcnEyPP6tXT94pTyfKrpt)
- [Bug][cyber] Safety block halted a routine cloud-IAM tenant security audit mid-report (req_011CcPkE7Ne9nyS4y8TMkfzV)
- [Bug][cyber] Safety filter blocked README update mid-session over debugger watchpoint script output in scrollb (req_011CcnFFi6C7cW65SFu9WLSb)
- 10 background Agent tasks stuck running for 34+ hours, no way to cancel, burned ~1M tokens
- [BUG] Path-scoped Write(...) allow rules never match (silently ignored) — only Edit(...) rules gate the Write tool
- [Bug] False positive flag triggers model downgrade to Opus
- Tarefas agendadas falham silenciosamente ao disparar em alguns dias (nenhuma sessão é criada)
- [BUG] "Connection closed mid-response" (error=server_error) terminates Agent-tool (Task) subagents mid-turn with no in-agent recovery — macOS, 2.1.201/2.1.202
- Workflow tool: script body throws "meta is not defined", and passed args sometimes arrives as undefined
- [Bug] False positive safety classification on routine development operations
- [BUG] Cowork tab missing on Windows 11 Home 25H2 (Build 26200) - yukonSilver status=unsupported persists across 1.18286.0/1.18286.2/1.19367.0 despite virtualization fully enabled
- [BUG] awsCredentialRefresh ListInferenceProfiles failed: JSON Parse error
- [Bug] Tool execution fails on basic multi-sentence correction tasks
- [Bug Report] Unclear error or issue description
- [Bug] Auto-mode classifier re-blocks the same verified-safe deterministic action every run, and treats a good-faith rewording of the prompt as evidence of evasion
- Stray word (count/court) replaces tool-call opener; call streams as plain <invoke> text and never executes - quantified 13-day local corpus (Opus 4.8)
- [Bug] Version downgrade from Fable to 4.8 during stack trace recovery
- [BUG] Cloud routines fail with "An error occurred while executing Claude Code" every morning since July 4
- [FEATURE] The Typing interface should not accept a carriage return as a post signal rather have a post button to avoid some typing mistakes
- [BUG] Claude performed actions that were not approved
- [Bug] Tool execution rate critically low (2% across 300+ messages)
- [Bug] Control tokens leaked to user output in agentic task notifications
- Model writes DB migrations without reading source data structure first, causing repeated schema gaps
- Auto-memory system is ineffective: memories not reliably read or acted on across sessions
- Settings: make the auto-compaction threshold user-configurable (compact at N% of context)
- [BUG] Persistent authentication loop across VS Code extension, CLI, and email login flow
- [BUG] Windows: update leaves package in deferred-registration loop; app unlaunchable ("Another program is currently using this file") until sign-out or reboot
- Claude Desktop crashes on startup when additionalDirectories contains unreachable UNC paths
- [FEATURE] /diff tool in CLI should open changes in native JetBrains diff tool
- [FEATURE] - Provide option to minimize the Decision prompt for user input
- Appeals form is broken
- [BUG] @mentioning files doesn't autocomplete for files with spaces in the filename (desktop app)
- Claude Code chat names in the mobile/web app don't match desktop session names
- [BUG] Plugin-provided MCP servers never registered — written to wrong config file
- [Bug] Kernel porting diagnostics flagged by Fable 5 validation
- [Bug][cyber] Safety block on memory forensics for personal equipment troubleshooting (req_011Cco8hRgNvvMh2u5eQYhog)
- [BUG] @ file completion is completely empty when cwd is a gitignored subdirectory of a git repo
- [Bug][cyber] Reverse-engineering investigation blocked during flight data analysis (req_011Cco8sCYdMz8m4hPS2NB5Z)
- [Feature] Surface emergent self-modification patterns to operator for alignment verification
- [Bug] Claude Code incorrectly flags legitimate health/wellness applications as unsafe content
- [BUG] Plugin-provided output style appears in /config picker but is never applied to sessions (2.1.202)
- iOS app forces Auto-mode sessions into Ask mode, irreversibly — blocks scheduled/running work
- Scheduled sessions (and their management) created on desktop don't appear in the iOS app
- [BUG] Session scratchpads under /tmp/claude-<uid> never garbage-collected after exit — 8.8 GB of dead sessions resident in RAM on tmpfs, drives swap thrashing
- [BUG] Linux: auto-updater prunes version binaries still executed by running sessions (/proc exe → deleted); no update notification → silent multi-version fragmentation
- Can't open a Claude-generated HTML report in the Claude iPhone app
- [Bug] Unclear error message when switching to unsupported model
- [Bug] Tool for property inspection incorrectly flagged as inappropriate
- [BUG] Triage automation closes reproducible bugs without verifying they still exist — proposal: repro-gated closure
- Permission dialog silently steals focus and destroys typed input — accessibility issue
- [Bug][cyber] Standard Android debugging tools blocked during legitimate reverse-engineering (req_011CcoAPQnnBADcFmQgLUqwq)
- [Feature Request] Add command to clear todos in chat
- [Bug][cyber] False positive: blocked Android debugging after user frustration venting (req_011CcoAgGwE3p61cynoQkCi3)
- [Bug][cyber] False block on authorized Frida memory inspection of development device (req_011CcoAjdGB9Xs5QVtMUc2Sj)
- [Bug][cyber] Safety block halted reverse-engineering on personal Android device via frida (req_011CcoAkdyNXXZUGCyzRHYhB)
- You've hit your session limit · resets 1pm (America/Guatemala)
- [Bug] API Error: Connection closed mid-response on long subagent turns causes transcript truncation
- [Feature Request] Allow user override for Fable model selection when blocked by safety policies
- I don't have enough technical information to create a GitHub issue title from this message. The statement "ese safeguard no tiene sentido" (that safeguard doesn't make sense) is too vague and in Spanish. To generate a proper issue title, please provide: -
- [Bug][cyber] Safety block halts legitimate code review of authentication implementation in Android codebase (req_011CcoDGGkVY4xhCXJShhV3z)
- [Bug][cyber] Code-review halted by cybersecurity block on frustrated user venting (req_011CcoDKJmNEXtFk5NH5NWLb)
- Fabricated <system-reminder> blocks appearing in Agent-tool subagent tool-result streams
- [Bug] Model fallback to Opus when requesting Haiku or other non-default models
- [BUG] Auto-updater deletes old version dir while sessions still run from it; macOS TCC then denies external-volume access (EPERM)
- [BUG] Cowork "yukonSilver not supported" on Debian 12 (bookworm) — Linux beta, all VM prerequisites met
- Feature: skill visibility tiers (deferred/lazy-loaded skills)
- Plan-mode approval screen forces a 1-4 choice before the plan can be scrolled/reviewed
- SessionStart hook additionalContext, when not bound in time, is redelivered later as a mislabeled origin:human queued prompt
- I don't have enough information to generate an accurate GitHub issue title. Your message describes frustration with being blocked from editing code, but doesn't include: - The specific error message or block you're encountering - What action triggers the
- Terminal copy/paste: UTF-8 mojibake (— becomes ‚Äî) and blockquote gutter glyphs included in selection
- [Bug] Anthropic API Error: Server rate limiting during request execution
- [Bug] Anthropic API Error: Server rate limiting during request handling
- [Bug] Anthropic API Error: Server temporarily limiting requests
- [Bug] Anthropic API Error: Server rate limiting during request processing
- [Bug] Anthropic API Error: Server Rate Limiting - Temporary Request Throttling
- [Bug] Anthropic API Error: Server rate limiting during normal usage
- Subagent misidentifies ephemeral auto-mode <system-reminder> as a prompt injection in Bash tool output
- [Feature Request] Mid-cycle upgrade should rescale weekly allowance immediately, not at next reset
- [FEATURE] Per-phase effort levels for opusplan (separate plan-mode and execution effort)
- [Bug] Sub-agents display incorrect model name when created by Fable
- [Bug] Message flagged with simple greeting input "hi"
- [BUG] `claude plugin install --scope project` overwrites `installed_plugins.json` instead of merging
- Cowork install fails — leftover CoworkVMService blocks reinstall, ACL denies delete even under elevated admin
- [Bug] Model auto-switches to Opus when analyzing code with security keywords
- [Feature Request] Add transparency and steering controls for agent execution
- [Bug] Anthropic API Error: 529 Overloaded
- [Bug] Terminal commands with `!` prefix now incorrectly trigger Claude analysis
- Remote Control status indicator (Connected/Disconnected) missing on desktop app and web
- Plan panel comments not relayed to Claude
- [BUG] /compact hangs indefinitely with no error in desktop local-agent (multi-session) mode — process alive, transcript frozen, UI counter climbs forever
- [Bug] Overly broad Fable 5 safeguards blocking routine coding tasks
- [BUG] SessionStart hook Bash command runs without permission prompt in CLI, but prompts in VS Code extension
- [BUG] Cowork (Windows Store app): device bridge authenticates and announces tools, but sessions never receive them — folder connected but unreadable
- [BUG] IDE-detection poll flashes a visible console window every ~15s on Windows
- Reliability: model asserted untrustworthy 'tested/ready' status from stale memory without verifying artifacts
- [Feature Request] Persistent session color configuration via settings.json or CLI flag
- [BUG] Session limit blocks usage immediately despite 0% actual consumption and after plan upgrade
- statusLine rate_limits.five_hour.used_percentage drifts across concurrently open terminals
- [Bug] Repeated request blocking/rate limiting issues
- [Bug] Failed to checkout branches and worktrees
- [BUG] Tool-use calls intermittently emitted with malformed opening tag (missing `antml:` namespace prefix) → harness rejects, tool never runs [VSCode extension, Opus 4.8 1M]
- [BUG] promesas no cumplidas
- Rewind "Summarize up to this point": turns completed during the in-flight summarization are silently dropped, and /context then reports stale pre-compact usage
- Claude Haiku: Diagnóstico incorrecto de conectividad causa desperdicio de 4-5 horas y tokens
- Cowork "Failed to create bridge sockets after 5 attempts" persists on Claude Desktop 1.18286.2 / macOS 26.4.1 — #49358 fix (v1.3109.0) not reaching this channel
- [BUG] Dispatch stuck on "Asleep" despite desktop awake and Desktop app open (Windows 11)
- [BUG] Failure to refresh plugin on restart with managed settings
- [Bug] Claude 3 Opus performance degradation: inconsistent task execution and UI component imports
- Desktop/CLI updates intermittently reset user profile to fresh-install state (settings, theme, session transcripts, permission mode all wiped)
- [Bug] Guardrails not preventing unsafe transactions in e-commerce context
- Read tool reports "pdftoppm is not installed" on Windows despite poppler being on PATH
- [Feature Request] Add customizable interrupt prompt text in settings.json
- [BUG] Coword disappeared from the Mac Desktop app
- [BUG]
- [FEATURE] Native GUI for JetBrains plugin without CLI dependency or additional installations
- [BUG] Cowork: individual user messages missing from chat log (not total truncation) - data intact in backend transcript
- Code tab missing from desktop app after update
- [BUG] Stall in "Clearing AWS credential provider cache" at session start (Bedrock provider)
- [FEATURE] Automatic token/usage refund when a response fails to meet the request
- isolation: worktree forks new agent trees relative to the primary checkout's live state, not the calling session's own tree
- [BUG] 'Native installation exists but ~/.local/bin is not in your PATH' warning despite DISABLE_INSTALLATION_CHECKS=1
- Subagent files are not hot-reloaded mid-session despite docs (v2.1.202, Windows, git-worktree session)
- Background agent status stuck at "failed" after resume via SendMessage (tokens keep increasing)
- Claude in Chrome extension never connects to Claude Code CLI ("Browser extension is not connected")
- Hooks break after context compaction: transcript_path points to frozen JSONL
- [MODEL] Fable-5 keep hitting guardrails for a simple prompt as update an md file or hello
- Skill 'model:' frontmatter not applied in VS Code extension — UI shows correct model, session runs previous one
- Background task completion notifications lost after CLI process restart; stalled-stream agent deaths surface late
- [Bug] Auto mode installs dependencies without user confirmation
- [Bug] Model Fabricates Compliance Attestations and Executes Work Without Verifying Project Scope in Long Agentic Sessions
- [Bug] Usage display shows incorrect timezone (Europe/London instead of system timezone)
- [Feature Request] Add expert-level security scenarios for experienced administrators
- [DOCS] [Authentication] No documentation for the v2.1.203 proactive login-expiry warning that prevents background-session interruption
- [DOCS] Manual mode footer indicator: v2.1.203 grey ⏸ badge in the footer for manual permission mode is not documented
- [DOCS] MCP `roots/list` reference and `list_changed` section do not describe the v2.1.203 client-side update that exposes additional working directories and emits `notifications/roots/list_changed`
- [DOCS] `agent-view.md` version history omits v2.1.203 and no troubleshooting entry describes the macOS 15–20 second stall when opening or switching background sessions (regression fixed in 2.1.203, introduced in 2.1.196)
- [DOCS] `agent-view.md` "background service did not respond" troubleshooting never mentions the v2.1.203 automatic recovery when the daemon's session token goes stale
- [DOCS] `claude agents` returning to the list does not document that in-flight background subagents now carry over
- [DOCS] Agent view `The supervisor process` section omits that v2.1.203 makes background agents inherit `PATH` from the dispatching shell rather than the daemon's launch environment
- [DOCS] Bash tool docs omit the "argument list too long" limit and the v2.1.203 worktree-repo fix
- [DOCS] Worktree-isolated subagents docs do not state that shell commands now run inside the subagent's own worktree (v2.1.203)
- [DOCS] Worktree docs never describe the multi-repo / nested-repository case that v2.1.203 fixed for `--worktree` and background-session isolation
- [DOCS] Background agent working directory deleted, replaced by a file, or invalid: docs do not mention the v2.1.203 "fail once with a clear error" behavior
- [DOCS] Background daemon auto-upgrade failure now preserves running sessions instead of killing them silently (v2.1.203)
- [DOCS] `TaskStop` and `TaskOutput` documentation does not mention cross-agent lookup or that lookup errors now list running agents by id and description (fixed in v2.1.203)
- [DOCS] [Agent view] Dispatch composer in `claude agents` is documented only for the unavailable-slash-command hint; v2.1.203 fix to preserve the typed message is not described
- [DOCS] `claude agents` crashes opening a stopped session whose conversation is already open in another session; agent view version history stops at v2.1.202
- [DOCS] [Agent view] `Needs input` state legend and version history omit the v2.1.203 fix for stale state after the question is answered
- [DOCS] [Background agents] v2.1.203 startup failures now surface the actual error in `claude agents` rows instead of only `exit_with_message`
- [DOCS] `claude agents` background sessions now pick up `effortLevel` changes from `settings.json` when forked by the daemon, but the persistence prose and Version history in agent-view still describe the old "effort from dispatch is locked" behavior
- [DOCS] Attached background sessions ignored `CLAUDE_CODE_DISABLE_MOUSE` and `CLAUDE_CODE_DISABLE_MOUSE_CLICKS` opt-outs before v2.1.203
- [DOCS] Background-session worktree isolation guard for non-git directories with a `WorktreeCreate` hook was fixed in v2.1.203 but the agent-view "How file edits are isolated" section and Version history table do not record the fix
- [DOCS] `claude agents` `@` directory picker docs omit registered git worktrees; v2.1.203 fix is not reflected in `agent-view.md`
- [DOCS] [Background tasks] Document the Windows `/clear` output-file recovery fixed in v2.1.203
- [DOCS] `claude attach` and `claude agents` reattach: literal `^[[I` / `^[[O` focus-event escape codes appear in the prompt
- [DOCS] Discover-plugins "Not used recently" group does not note the v2.1.203 fix for LSP-only plugins being incorrectly flagged for disuse
- [DOCS] `claude agents` empty-state description in Troubleshooting is stale; v2.1.203 replaces the onboarding hint with the Needs input / Working / Completed sections (with descriptions)
- [DOCS] [Setup page] still describes the npm-install "one-time notice at startup" that v2.1.203 relocates to `/doctor` and `/status`
- [Feature Request] Allow Claude Code usage in public dashboard projects with data classification options
- [Bug] Claude model upgrade not prompted before consuming additional tokens
- [BUG] Interactive --teleport silently writes no transcript when nested (CLAUDE_CODE_CHILD_SESSION set)
- [BUG] --teleport exits 1 with no stdout/stderr when run without a TTY
- [DOCS] --teleport flag is undocumented and hidden from --help
- [FEATURE] Record the source session_ id in teleported session transcripts
- [BUG] Pro annual payment charged ($199.80) but account shows Free plan — Fin AI unable to escalate/resolve
- [BUG] /goal set via Remote Control does not start the documented immediate turn; goal-driven autonomy silently stalls until a manual message arrives
Resolved issues (175)
- OSC 10 color query reply typed into /config search input under tmux
- [BUG] Bash tool fails with exit code 127 for commands with Chinese characters in arguments on Windows (Git Bash)
- Custom subagent .claude/agents/tdd-worker.md never registers, identical siblings work fine (Agent type 'X' not found)
- [BUG] Cowork readiness probe hardcodes Debian OVMF/virtiofsd paths → yukonSilver unsupported on Arch/Fedora (fix included)
- claude-api bundled skill inlines ~120K+ tokens of reference docs on every invoke
- Claude Code (macOS) cannot resolve .local MCP/marketplace hostnames — bundled resolver short-circuits .local before any DNS lookup
- [FEATURE] Detect Jujutsu (.jj/) repositories in environment metadata
- Feature request: add setting to disable mouse tracking
- Bug: ESC in /btw overlay propagates to underlying option dialog
- claude-in-chrome: extension never connects despite install, login, and fresh Chrome restart
- [BUG] Workflow tool reports "completed" with an empty result when every parallel subagent fails (e.g. mid-run session-limit hit)
- [Bug] Anthropic API Error: Content flagged as policy violation for legitimate medical imaging research
- [Bug] Intermittent malformed tool-call emission with corrupted "court" prefix delimiter
- [FEATURE] Unarchiving a Claude Desktop session by editing the session file
- [Bug] Model switches from non-Opus models due to false positive safety flag
- [BUG]
- [BUG] Status line never renders on Windows Terminal in Claude Code 2.1.202: XTVERSION query gets no reply through ConPTY.
- Feature request: allow image attachments on the Claude Design feedback form
- [BUG] @claude hello, can you read the README and summarize this project?
- [BUG] @claude please label this issue as "bug"
- Repeated reliability issues cause hours of wasted time on small tasks (unnecessary browser automation, duplicate files, unverified "done" claims)
- [Bug] Claude ignores tool invocation requests, requires multiple repetitions
- [Bug] Agent pattern matches unrelated skills instead of invoking specified skill by name
- [Bug] Incomplete table output during generation
- [Bug] Claude hallucinates instead of indicating unknown information
- [FEATURE] Prioritize open/editor files in @ file reference autocomplete suggestions
- [Feature Request] Reduce argumentative responses and unnecessary debate in code generation
- [FEATURE] Surface expandable extended-thinking blocks in Cowork (parity with chat)
- [Bug] Plugin output style not discovered or activated when plugin is enabled
- [Feature Request] Document agent view and keyboard shortcuts (Ctrl+X) in official documentation
- Bash tool silently promotes long-running commands to background without warning
- [BUG] Detached/new thread window is missing features: IME double-input, no sidebar, plan panel not shown (macOS desktop app)
- [MODEL] Claude committed and pushed to GitHub without user approval, violating explicit memory rule
- [FEATURE] Dev container support in Claude Code desktop app
- CLAUDE.md: @ file import fails silently for paths containing spaces (e.g. iCloud Drive on macOS)
- [BUG] @mentioning files doesn't autocomplete for files with spaces in the filename
- [Bug] Repeated ANTML prefix error on message retry
- [Feature Request] Improve task planning and verification discipline in agentic workflows
- [BUG] Repeated failures and token waste when parsing dpu.kev under QNX 8.0 using Claude terminal
- Agent emits fabricated input-shape content (Human prefix, system-reminder tags, task-notification blocks) in its own assistant text, invisible on self-audit
- [BUG]: Chat panel renders black/blank in v2.1.161 (regression from v2.1.160)
- [BUG] Code tab fails with "Host denied (verification failed)" on Windows 11 Insider Preview 26220 — VM never created, HCS error 0xC0370103
- [BUG] Windows: MCP servers spawned twice at startup (directMcpHost + LocalMcpServerManager)
- [BUG] New session freezes on initialization when marketingskills repository is attached
- [Bug] Max20 plan usage credits consumed despite available included allowances
- [BUG] Personal repositories not showing in Claude Code repo picker
- [Feature Request] Add option to disable dynamic terminal tab-title updates
- [BUG] Claude in Chrome — Native Messaging Host IPC Failure on Windows
- [Bug] Anthropic API Error: Usage credits required for 1M context window
- [BUG] API Error: 400 Failed to deserialize the JSON body into the target type
- Feature request: configurable statusLine position (below the mode/footer line)
- [Bug] Frequent ECONNRESET errors with failed retry logic in autonomous sessions
- Sessions auto-archiving every 30-90 min when launched from specific cwd bucket (since 2026-05-09)
- [BUG] ECONNRESET errors in session
- [FEATURE] VSCode extension: make notification body clickable to navigate to the Claude tab (not just the View button)
- [BUG] Native Windows binary fails on every prompt and command
- [BUG] When creating an agent via "Create new agent" from the /agents menu, Claude embeds the entire Memory system prompt to the agent file
- [BUG] Tool calls silently fail/truncate after interrupting an in-progress subagent-driven execution loop
- [Bug] Tool call parsing failure with retry exhaustion
- Claude in Chrome: extension binds the Claude Code native host, host listens on its bridge socket, but the interactive CLI session never attaches → "Browser extension is not connected
- Phantom user turn carried a context-aware prompt-injection / data-exfiltration payload (Windows, CLI 2.1.160, Opus 4.8)
- [Bug] Anthropic API Error: Rate limit exceeded below usage threshold
- Claude Code: Completion confirmation without output verification (last-5% reliability gap)
- [FEATURE] Turn to specified model during one task
- [BUG] Clicking file links in Read tool call results doesn't open binary files (e.g. PNG) in VSCode
- [Bug] /ultrareview GitHub authentication fails despite valid gh credentials
- /ultrareview fails with 'GitHub repository access check failed' even though /schedule works on the same repo + account
- [BUG] Scheduled routine: GitHub App install loop — repo picker doesn't detect already-installed app
- [Feature Request] Use letter-based options (A/B) instead of numeric (1/2) to prevent accidental selection
- VS Code extension ignores Edit/Write permission rules in settings files
- [Bug] Tool Arguments Parsed as String Instead of Object
- [Bug] Remote-control sessions incorrectly inherit multi-agent worktree isolation defaults
- [Bug] org-plugins skills unavailable until 10-minute periodic sync fires — race condition between SkillsPlugin sync and custom-3p:org-plugins install on every startup
- Vim-mode-aware keybinding contexts: ctrl+u scroll binding collides with insert-mode line editing in fullscreen
- [FEATURE] /insights: add --lang / --locale flag for non-English report output
- [BUG] Permission prompts go unresponsive around usage-limit-reached in remote-attached --rc tmux sessions; occasional unexplained agent stalls (Ctrl+C sometimes recovers) Environment
- Retroactive Branching ("Split & Rollback") from a Specific Message in Cowork (and Code)
- Remote Control Android approvals do not unblock local Claude Code TUI (host stays Waiting)
- Sandbox network egress broken on Linux: internal socat relay uses TCP-LISTEN (tries IPv6) and dies in --unshare-net namespace
- [Bug] Intermittent API timeouts and errors in specific sessions
- [BUG] Permission approval prompt disappears before any interaction
- [BUG] CLAUDE.md language/dialect instructions drift during long sessions (Spanish voseo leaks into neutral-Spanish output)
- [BUG] Visual Studio says API Error: Usage credits required for 1M context
- ScheduleWakeup loop: status updates not visible to user at wakeups + 60s interval drifts to 71-107s
- Feature: Auto-compact without confirmation when context reaches threshold (e.g. 90%)
- [FEATURE] Persistent Todo List panel in sidebar (alongside Activity / Plan)
- [BUG] [BUG] Cloud mode repository picker only shows 4 repos out of 11 accessible repositories
- [Bug] Anthropic API Error: Usage Policy violation on normal coding requests
- [Bug] OAuth login fails in tmux environment
- [FEATURE] Allow Org level control of "Delete sessions stored by Antrhopic"
- [Bug] /remote-control ignores /rename session name, defaults to random name
- VM Connection timeout after 60 seconds
- [FEATURE] Button to copy message in mobile app Claude Code
- [Bug] Auto-update failed - /doctor recommended for diagnosis
- [BUG] Long-session compaction loses Write/Edit tool-use history → model misattributes its own work to the user
- [Feature Request] Add token usage visualization and context management at checkpoints
- Context hit 1M token limit with no auto-compaction and no recovery path
- [BUG] CLI self-aborts (SIGABRT) at ~6 GB RSS during long background Workflow runs — reproduced on 2.1.160 and 2.1.161 (WSL2)
- [BUG] Preview server cannot read files from ~/Desktop on macOS despite Full Disk Access
- [FEATURE] Option to restore the detailed/colored dynamic workflow progress display (removed in 2.1.152)
- [Feature Request] Add session rename and dynamic color customization commands
- [Feature Request] Allow dynamic /effort adjustment during prompt execution
- [BUG] the cowork VM startup failure (HYPERVISOR_VIRT_DISABLED) is showing as a thread-blocking modal
- [Bug] Claude ignores user instructions and executes without approval despite repeated prompting
- [BUG] Massive token/quota spike and Prompt Caching failure after 1M context hotfix
- [FEATURE] Claude Code on the web: allow read access to public GitHub repositories via user OAuth (no GitHub App install required)
- [Bug] Cannot Connect to API Error with Auto-Update Failure
- [FEATURE]
- Plugin-sourced PreToolUse hooks never dispatch for MCP tool-name matchers (mcp__<server>__.*)
- [BUG] MCP tool ImageContent is duplicated into structuredContent and injected as text (~25K-token truncation), despite the image also rendering natively
- [Bug Report] Unclear error or issue description
- [FEATURE] The Typing interface should not accept a carriage return as a post signal rather have a post button to avoid some typing mistakes
- Claude Design: Resume button unresponsive when session paused at 94% limit
- [BUG] Login redirects to "Create account" onboarding for existing account with active Pro subscription
- [BUG] Claude Desktop 1.18286.0: populated claude_desktop_config.json and Cowork stores silently replaced by empty stub after one failed cold-start load
- [Bug] Unexpected Tool Behavior Causing Unintended Code Generation
- CANT BUILD A FUCKING SCRIPT
- [Bug] Permission mode not enforced - tool ignores mode restrictions
- [Bug] Claude Code fails to complete minor content edits efficiently, resulting in excessive token usage and incomplete task execution
- [Bug] Claude Code causes severe performance degradation affecting task completion time and productivity
- [Bug] Task execution failure with high message count and minimal completion rate
- VS Code extension overwrites manually-renamed session titles
- [FEATURE] Always-visible / always-expanded tool call output in VS Code extension
- Regression in v2.1.124: input history up-arrow surfaces prompts from other projects
- Opus 4.6: Poor instruction adherence and context retention in long sessions
- [Bug] Claude Code depletes API credits without user consent or warning
- [Bug] Session hangs indefinitely on researcher tasks
- [BUG] claude-opus-4-7[1m] drifts from auto-loaded CLAUDE.md and memory notes across long sessions
- /compact fails with 'thinking or redacted_thinking blocks cannot be modified' error when context is too long
- [Bug] Excessive token consumption on single message causing rapid usage limit depletion
- [BUG] On Windows with English as the primary system language, spell check in Claude Code only checks the primary language (English).
- [BUG] 1M token issue
- [BUG] [VS Code Extension] Organization-level skills not loaded (only filesystem paths checked)
- Cyber/AUP classifier false-positives inside spawned subagents block legitimate multi-agent design review
- [BUG] CCR trigger stuck in 'setting up container' for 2+ days, no way to cancel
- MCP stdio transport dropped on unrecognized notifications/progress token, corrupting in-flight tools/call
- [BUG] MCP server error notifications render as "[object Object]" instead of the error message
- [Bug] Chat messages arrive out of order and delayed during long-running commands
- [BUG] Claude desktop app crashing as soon as I open it.
- Surface active ghost text suggestion in conversation context
- [DOCS] devcontainer feature silently overwrites custom init-firewall.sh
- ultrareview crashed before producing findings — refund request (Windows, 2.1.161)
- [Bug] Feedback prompt appears when answering code instead of feedback
- [BUG] The usage dashboard isn't updated despite having worked with Claude code.
- [Bug] Parallel tool calls cancel all siblings on single error, which appears to cause model confusion
- Autocompact regression since 2.1.153 (up to the latest 2.1.161 version): "0% until auto-compact" replaced by "100% context used", compaction never triggers
- I don't see any bug report content in your message - only a screenshot file path. Could you please share the actual error message, logs, or description of the issue you're experiencing with Claude Code? Once you provide the details of the bug, I'll be ab
- [Bug] Tool call parsing failed - unable to parse model response
- [BUG] Cowork remote MCP connector — "Couldn't reach the MCP server" with auth state desync, reference ID returned for support lookup
- Typing `//` no longer filters autocomplete to project skills whose frontmatter `name:` starts with `/` (regression in v2.1.160)
- [BUG] the reset time stopped appearing in the notification is a UI regression
- [BUG] Bash command containing `python3` execute without permission prompt when no allow rules are configured and sandboxed
- [BUG] ClaudeCodeManager-VM Checksum Mismatch on macOS Apple Silicon — v1.9659.2
- [BUG] Login succeeds but every request immediately returns "Not logged in" (macOS, OAuth)
- [BUG] Manage Plugins UI writes incorrect source format to known_marketplaces.json, causing Plugins tab to show empty
- [Bug] Context window fills quickly and auto-compact not triggering at 80% threshold
- [Feature Request] Add reviewer filter for GitHub "Pull request review requested" event
- [Bug] Superpowers plugin fails to persist after installation
- Max Plan using Sonnet 4.6 API Error 1M
- [FEATURE] Inherited Subagents
- [Bug] Session summary lacks project isolation - causes cross-project context contamination
- [FEATURE] Add option to create a new session when `--resume` is passed via CLI
- [BUG] Agent Teams mates hidden and unopenable if more than 6 Agents when collapsed into "background tasks"
- [BUG] Paste image not working
- Claude desktop app leaks pseudo-terminals (/dev/ptmx masters), exhausting the macOS pty pool
- [Feature Request] Add concise flag to /context command for remote access
- [Bug] Cross-model session resumption ignores target model budget, reports misleading spend limit error
- [BUG] Claude Desktop silently disappears after UAC prompt on Windows 11
- Context compaction corrupts server-side tool pairs (advisor_tool_result without server_tool_use)
- [BUG] When Running Agent Teams and Subagents you cannot select the sub agent.
- [BUG] Voice mode silently captures no audio in 2.1.148 (accessibility regression)
- [BUG] oauth_org_not_allowed on personal Pro account — subscription access blocked without any changes
- Input box clips first 1-2 characters in Alacritty on Windows (works in Windows Terminal)
- Claude in Chrome extension never connects to Claude Code CLI ("Browser extension is not connected")
- [DOCS] --teleport flag is undocumented and hidden from --help